.... l . i... 1th“ c. si.éfx,l;l—Y 2...... ii 9 ”Guyana.“ 0).". 5511!: I. 33...! . XII-.1 {vii 110)]: e!!!. C? r «Al... lv:.v(.. V. .t....’:.. . . ... .2... ‘...~Lo.. Jellll.rbnléro..tlr V , . Clot .puoul.¢la 2| nth.".l. 61,. . .a : . . .. . 1.7.." . . 99:... It: . v), a; z . n :1. o'. . 7;..Il, I: I: . 9 . . . .... I. . . .. V. -sl:v. ft‘. , ‘1 .V- -.t .n. ;! .. 41‘..- Lil. 'Il' l TTE UNIVERSITY LIBRARIES IIIIIIIIIIIIIIIIII II IIIIIII II III III 31293 008 This is to certify that the dissertation entitled The Effect of Environmental Factors on the Function and Structure of Porins from Escherichia coli K-12 presented by Jill Cokeen Todt has been accepted towards fulfillment of the requirements for 421341—— degree in Jiochemjstry Date l /“/q-3 MS U is an Affirmative Action/Equal Opportunity Institution O-IZTI‘I Mlc LIBRARY Untverslty hIgan State PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE II @QMWT I I IE [,1ng II::I VIII II‘”II I MSU I: An Affirmative Action/Equal Opportunity Immitution THE EFFECT OF ENVIRONMENTAL FACTORS ON THE FUNCTION AND STRUCTURE OF PORINS FROM ESCHERICHIA COLI R-12 BY Jill Cokeen Todt A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Biochemistry 1993 ABSTRACT THE EFFECT OF ENVIRONMENTAL FACTORS ON THE FUNCTION AND STRUCTURE OF PORINS FROM ESCHERICHIA COLI K-12 BY Jill Cokeen Todt In this thesis, we present evidence for an active role for porins in the regulation of the permeability of the outer membrane of Escherichia coli K-12. In particular, we have detected a pH-induced switch in channel size in vitro and in vivo with porins OmpF, OmpC, and PhoE. Small channels were detected at acidic pH with a transition to a set of larger-size channels (~double the cross-sectional area) at basic pH. This pH-induced switch between porin conformations occurred near physiological pH and appeared to be reversible. The pKa for this transition suggested that the only histidine, conserved in all three porins, may be involved in the pH-induced switch in channel size. Further evidence for the involvement of H1521 came from studies showing that modification of H1821 with diethyl pyrocarbonate eliminated the pH-induced switch in channel size (large-size channels were present at all pH's). Other factors besides pH were also found to affect channel size in vitro. These included the amount of LPS bound to porin, voltage, the presence of M00, and the tension of the membrane into which porin inserts. Some of these factors could be responsible for short-term regulation of channel size in vivo. The functional changes in porin were correlated to structural alterations with pH measured using Fourier transformed infrared spectroscopy and intrinsic tryptophan fluorescence. This correlation was evidenced by: 1) the similarity of the pKa's of the structural change and of the channel size transitions, and 2) the induction of a change in structure by an alteration in pH similar to that induced by the modification of Hile. Based on this data, we propose that at acidic pH, Hile (which is across from the loop L3 between B strand 5 and 6 that defines the channel exclusion limit) becomes protonated causing L3 to come closer which narrows the channel. This mechanism, in combination with regulation of channel size by the synthesis of different levels of porins (OmpF is produced in high amounts at basic pH, while OmpC is produced at acidic pH), could protect the infectious cell from antibiotics or host defense mechanisms since fluid from sites of bacterial infection in humans has been found to be acidic. This is dedicated to my parents and my husband for their unconditional love and support. ACKNOWLEDGEMENTS I would like to express my deepest gratitude to the following people: My thesis adviser Dr. Estelle McGroarty for her patience, tolerance, and good humor as well as her excellent guidance. Mildred Rivera, Warren Rocque, and especiaLly Barb Hamel for their great friendship, support and advise. The members of my thesis committee, Dr. Shelagh Ferguson- Miller, Dr. Alfred Haug, Dr. Rawle Hollingsworth, and Dr. Jack Holland for their help and interest. TABLE OF CONTENTS Page List of Tables . . . . . . . . . . . . . . . . . . . x List of Figures . . . . . . . . . . . . . . . . . . xi List of Abbreviations . . . . . . . . . . . . . . . xiv Chapter 1 Literature Review . . . . . . . . . . . . . . 1 Gram-negative Bacterial Cell Envelope Structure . . . . . . . . . . . . . . . . . . 2 Escherichia coli Porins and Regulation of Synthesis . . . . . . . . . . . . . . . . . . 5 Bacterial Porin: Structure . . . . . . . . . 8 Bacterial Porin: In vitro Functional Analyses 15 Bacterial Porin: In vivo Functional Analyses 21 Bacterial Porin: Structure-Function Studies . 23 Bacterial Porin: Significance of Structure- Function Studies . . . . . . . . . . . . . . . 24 Chapter 2 Effects of pH on Bacterial Porin Function . . 31 Abstract . . . . . . . . . . . . . . . . . . . . 32 Introduction . . . . . . . . . . . . . . . . . . 33 Materials and Methods . . . . . . . . . . . . . 39 Cell Growth . . . . . . . . . . . . . . . . . 39 Porin Isolation . . . . . . . . . . . . . . . 39 Bilayer Lipid Membrane Assay . . . . . . . . . 41 Liposome Swelling Assay . . . . . . . . . . . 42 Results . . . . . . . . . . . . . . . . . . . . 43 pH Effects on Channel Size . . . . . . . . . . 43 Effects of Voltage and LPS Depletion . . . . . 54 Discussion . . . . . . . . . . . . . . . . . . . 69 vi Page References . . . . . . . . . . . . . . . . . . . 76 Chapter 3 Involvement of Hile in the pH-induced Switch in Porin Channel Size . . . . . . . . . . 80 Abstract . . . . . . . . . . . . . . . . . . . . 81 Introduction . . . . . . . . . . . . . . . . . . 82 Materials and Methods . . . . . . . . . . . . . 83 Cell Growth . . . . . . . . . . . . . . . . . 83 Porin Isolation . . . . . . . . . . . . . . . 83 Carbethoxylation and Decarbethoxylation of Porin O O O O O O O O O O O O O O O O O 84 Bilayer Lipid Membrane Assay . . . . . . . . . 85 FTIR . . . . . . . . . . . . . ... . . . . . . 86 Results . . . . . . . . . . . . . . . . . . . . 88 Discussion . . . . . . . . . . . . . . . . . . . 96 References . . . . . . . . . . . . . . . . . . . 99 Chapter 4 Effects of pH on Bacterial Porin Structure; the Involvement of Hile in a pH-induced Conformational Change . . . . . . . . . . . . 101 Abstract . . . . . . . . . . . . . . . . . . . 102 Introduction . . . . . . . . . . . . . . . . . 103 Materials and Methods . . . . . . . . . . . . 104 Cell Growth . . . . . . . . . . . . . . . . 104 Porin Isolation“ . . . . . . . . . . . . . . 104 Carbethoxylation and Decarbethoxylation of Porin . . . . . . . . . . . . . . . . 105 FTIR . . . . . . . . . . . . . . . . . . . . 106 Intrinsic Tryptophan Fluorescence . . . . . 108 Results . . . . . . . . . . . . . . . . . . . 109 vii Page Discussion . . . . . . . . . . . . . . . . 120 References . . . . . . . . . . . . . . . . . 124 Chapter 5 Factors Affecting the pH-induced Switch in Channel Size of OmpF and OmpC . . . . . . . 126 Abstract . . . . . . . . . . . . . . . . . . . 127 Introduction . . . . . . . . . . . . . . . . . 128 Materials and Methods . . . . . . . . . . . . 130 Cell Growth . . . . . . . . . . . . . . . . 130 Porin Isolation . . . . . . . . . . . . . . 130 MDO Isolation . . . . . . . . . . . . . . .i 131 Bilayer Lipid Membrane (BLM) Assay . . . . . 131 Results and Discussion . . . . . . . . . . . . 133 References . . . . . . . . . . . . . . . . . . 142 Chapter 6 Analysis of Mutant OmpF Porins from Escherichia coli K-12 strains 0C905, OC904, and 0C901 . . . . . . . . . . . . . . . . . . 143 Abstract . . . . . . . . . . . . . . . . . . . 144 Introduction . . . . . . . . . . . . . . . . . 145 Materials and Methods . . . . . . . . . . . . 147 Cell Growth and Porin Isolation . . . . . . 147 Bilayer Lipid Membrane (BLM) Assay . . . . . 147 Results . . . . . . . . . . . . . . . . . . . 149 Discussion . . . . . . . . . . . . . . . . . . 156 References . . . . . . . . . . . . . . . . . . 158 Chapter 7 Acid pH Decreases OmpF and OmpC Channel Size in Vivo . . . . . . . . . . . . . . . . . 159 Abstract . . . . . . . . . . . . . . . . . . . 160 Page Introduction . . . . . . . . . . . . . . . . . 161 Materials and Methods . . . . . . . . . . . . 163 Cell Growth . . . . . . . . . . . . . . . . 163 Assay for Cephalosporin Hydrolysis . . . . . 163 Crude fl-lactamase Preparation . . . . . . . 164 Results and Discussion . . . . . . . . . . . . 165 References . . . . . . . . . . . . . . . . . . 170 Chapter 8 Summary and Perspectives . . . . . . . . 172 ix LIST OF TABLES Page Chapter 2 1. Liposome swelling assay . . . . . . . . . . . . . . 55 2. Voltage gating of OmpF . . . . . . . . . . . . . . 61 3. Voltage gating of OmpC . . . . . . . . . . . . . . 62 Chapter 6 1. Voltage gating of wild-type and mutant OmpF porins 154 Chapter 7 1. Rate of hydrolysis of antibiotic by cells producing either OmpF or OmpC and by crude B- lactamase isolates . . . . . . . . . . . . . . 166 LIST OF FIGURES Page Chapter 1 1. Schematic representation of the cell envelope of Escherichia coli . . . . . . . . . . . . . . . . . 4 2. Tentative structure of E. coli K-12 LPS. . . . . . 7 3. Topology of OmpF . . . . . . . . . . . . . . . . . 10 4. RIBBON diagram of OmpF . . . . . . . . . . . . . . 13 Chapter 2 1. Stepwise current changes across a membrane comprised of phosphatidyl choline in the presence of LPS- depleted OmpF in a bathing solution of 0.5 M NaCl and 0.5 mM succinate (pH 5.4). . . . . . . . . . . 45 2. Probability distribution histograms of the size parameter k/o (A), for LPS-enriched wild type OmpF (A), OmpC (B), and PhoE (C) as measured in bilayer lipid membranes. . . . . . . . . . . . . . . . . . 47 3. Average size (h/o) of the large channel at pH's above 6.5 for LPS-enriched OmpF (A) and OmpC (B). . 50 4. Titration curve for the pH-induced switch in channel size for OmpC (A), and OmpF (8), based on bilayer lipid membrane assay analysis performed as described in Figure 2. . . . . . . . . . . . . . . . . . . . 53 5. Probability distribution histograms of the size parameter A/a (A), for LPS-enriched and LPS-depleted OmpF as measured in bilayer lipid membranes. . . . 57 6. Probability distribution histograms of the size parameter X/o (A), for LPS-enriched and LPS-depleted OmpC as measured in bilayer lipid membranes. . . . 59 7. Analysis of Voltage ating for LPS-enriched OmpF at pH 5.4 (0) and 9.2 ( ), and for LPS-enriched OmpC at pH5.6(I)and9.2(CI)............... 65 8. The effect of voltage on channel current for LPS- enriched OmpF (A) and OmpC (B) at pH 5.6 and pH 9.2. 67 xi Page Chapter 3 1. Probability distribution histograms of the size parameter A/o (A), for OmpF (A), DEPC-modified OmpF (B) and DEPC-modified OmpF treated with 20 mM hydroxylamine (C) as measured in bilayer lipid membranes. . . . . . . . . . . . . . . . . . . . . 90 2. Probability distribution histograms of the size parameter A/a (A) for OmpC (A), DEPC-modified OmpC (B), and DEPC-modified OmpC treated with 20 mM hydroxylamine (C) as measured in bilayer lipid membranes. . . . . . . . . . . . . . . . . . . . . 92 3. Amide I region of the FTIR spectra of LPS-enriched OmpF (A) or OmpC (B) suspended in 1% SDS and either 10 mM succinate, pH 5.8 (---) or 10 mM CHES, pH 9.25 (__). . . . . . . . . . . . . . . . . . . . . . . . 95 Chapter 4 1. Amide II regions of the FTIR spectra for OmpF (A)or OmpC (B) suspended in 0.1% SDS and either 10 mM CHES, pH 8.5 or 10 mM phosphate, pH 5.6. . . . . 112 2. Amide II regions of the FTIR spectra for either OmpF (A) or OmpC (B) suspended in 0.5% SDS, 10 mM phosphate pH 5.6 and treated with DEPC (a) or ethanol (b). . . . . . . . . . . . . . . . . . . 114 3. Quantum efficiency of the fluorescence emission spectra (Au=295 nm) for OmpF (A), PhoE (B), OmpC (C), and tryptophan and tyrosine at 40 uM and 290 uM respectively (D) . . . . . . . . . . . . . . . . 117 4. Quantum efficiency at peak maximum of the fluorescence emission spectra (k“=295 nm) for OmpF plotted against the pH of the buffer . . . . . . 119 Chapter 5 1. Effect of MDO on the percentage of large-sized OmpF channels based on bilayer lipid membrane analysis performed as described in Figure 3 . . . . . . . 135 2. Effect of MDO on the percentage of large-sized OmpC channels based on bilayer lipid membrane analysis performed as described in Figure 3 . . . . . . . ' 137 xii Page 3. Probability distribution histograms of the size parameter h/o (A) for OmpF and OmpC as measured in bilayer lipid membranes. . . . . . . . . . . . . 140 'Chapter 6 1. Probability distribution histograms of the size parameter A/o (A) for 0C901 OmpF (A), wild type OmpF (B), OC904 OmpF (C), and OC905 OmpF (D), as measured in bilayer lipid membranes at 25 mV. . . . . . . 151 2. Probability distribution histograms of the size parameter A/o (A) for 0C905 OmpF (A), wild type OmpF (B), 0C901 OmpF (C), and OC904 OmpF (D) as measured in bilayer lipid membranes at 75 mV. ._ 153 xiii BCA BLM CHES DEPC DSC EDTA FTIR KDO kDa LPS LSA MIC Omp PAGE SDS Tris VDAC LIST OF ABBREVIATIONS Bicinchoninic acid Bilayer lipid membrane 2-(N-cyclohexylamino)ethane Sulfonic acid Diethyl pyrocarbonate Differential scanning calorimetry Ethylenediaminetetraacetate Fourier transformed infrared spectroscopy 2-keto-3-deoxyoctulosonic acid Kilodaltons Lipopolysaccharide Liposome swelling assay Minimum inhibitory concentration Membrane-derived oligosaccharide Outer membrane protein Polyacrylamide gel electrophoresis Sodium dodecyl sulfate Tris(hydroxymethyl)aminomethane Voltage-dependent anion channel xiv CHAPTER 1 Literature Review 2 Gran Negative Bacterial Cell Envelope Structure In gram-negative bacteria, the cell envelope acts as a selective barrier allowing in nutrients and precluding harmful agents such as antibiotics, proteases, or bile salts. The selectivity is determined by the structure and composition of the cell envelope (Figure 1) and has been discussed in many recent reviews (Nakae, 1986; Lugtenberg & Alphen, 1983; Nikaido & Vaara, 1985). The cell envelope is formed by an outer and an inner membrane. A periplasmic space located between the two membranes contains the peptidoglycan layer. The periplasm also contains hydrolases, nutrient- binding proteins, and membrane-derived oligosaccharides (MDOs). MDOs are negatively charged polymers of glucose (8- 10 residues) with sn-1-phosphoglycerol, phosphoethanolamine, or O-succinyl ester residues substituted at various positions (Kennedy, 1982). MDO production is increased in cultures under conditions of low osmolarity (Van Golde et. al, 1973; Kennedy, 1982; Sen et. al., 1988) presumably to keep the periplasm isoosmolar with the cytoplasm preventing the cytoplasmic membrane from swelling significantly under these conditions (Stock et. al., 1977). The presence of fixed anions in the periplasm creates a Donnan potential across the outer membrane of 30 mV (Sen et. al., 1988). The outer membrane contains lipopolysaccharide (LPS) in the outer leaflet, phospholipid in the inner leaflet, and Figure 1. Schematic representation of the cell envelope of Escherichia coli. OM represents the outer membrane, PG=peptidoglycan layer, CM=cytoplasmic membrane. Lipopolysaccharide is represented by a, the outer membrane structural protein OmpA is represented by b, the outer membrane porins=c, phospholipids=d, and lipoprotein=e. 413. III M III] III III; IIIIIIII. IIIIIIIIIIIIIIIIIIIIIIIIIQ IIIIIIIIIIIIIIIInnIIIImIII fifififim: IIIIIIIIIIIQIIIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII IIIIIIIIIIII FIGURE I 5 a number of major and minor proteins. LPS consists of lipid A, core polysaccharide and an O-antigen (Lugtenberg & Alphen, 1983). Lipid A consists of 3 1,6 linked D- glucosamine disaccharides head groups covalently bound with fatty acyl chains, two of which are amide bound. Lipid A is essential for the endotoxin activity of LPS. The LPS core oligosaccharide contains 3-deoxy-D-manno-octulosonic acid (KDO) and a number of sugars such as glucose, galactose, and N-acetylfD-glucosamine. The O-antigen, used in O-serotyping to identify substrains of one species, consists of 40 or more repeating units each containing 3-6 sugar residues. These sugars can be substituted with O-acetyl groups, phosphate, amino acids, or ethanolamine triphosphate; this can contribute significantly to the net surface charge of the bacterial cell. In some strains, such as E. coli K12, LPS is missing the long O-antigen (Figure 2). EScherichia coli Porins and Regulation of Synthesis Nutrients enter the cell through outer membrane proteins called porins. A number of reviews (Benz 8 Bauer, 1988; Benz, 1985; Nikaido, 1992; Rosenbusch, 1990; Nikaido & Vaara, 1985; Nakae, 1986; Tommassen, 1988) have described porin structure and function; these proteins from enteric species form nonspecific water-filled pores (in the outer membrane) with a diffusion limit of ~600 Daltons. The two main nonspecific porins of Escherichia coli are OmpC and OmpF. OmpC has a smaller channel than OmpF and is produced Figure 2. Tentative structure of Escherichia coli K-12 LPS. Ions and other non-covalently attached constituents are not included. Amide and ester bonds are indicated by -N- and -0- links respectively. The following abbreviations are used: Glc=glucose, Gal=galactose, Hep=heptose, KDO=2-keto- 3-deoxyoctulosonic acid, Rha=rhamnose, GlcN=N-acetylglucosamine, EA=ethanolamine, P=phosphate. I. GIcNAc- -'-6>G|c lion: lacme3 -—'> Heir); h 6 ~ Gal Figure 2 8 under conditions of high osmolarity, low cAMP concentration, high temperature (Nakae, 1986), or low pH (Heyde & Portalier, 1987). The production of PhoE is also reduced under conditions of high osmolarity (Meyer et. al., 1990). PhoE, OmpF, and OmpC are defined as general diffusion pores since the movement of ions in their channels is similar to that in free solution; however, PhoE (induced by phosphate limitation) has binding sites for anionic solutes which makes it an anion—selective channel (Brunen et. al., 1992). OmpC and OmpF are slightly cation-selective (Benz et. al., 1979). LamB, whose production is activated by maltose or maltodextrin-containing media, is a specific porin since it. shows saturation behavior with increasing substrate concentrations (Schulein & Benz, 1990). Bacterial Porin: structure Because porins play such an important role in nutrient uptake, their structure has been examined extensively (Jap et. al., 1990; Chang et. al., 1985; Jap et. al., 1991; Dorset et. al., 1984; Cowan et. al., 1992; Weiss et. al., 1991; Fourel et. al., 1992; Jeanteur et. al., 1991). The primary sequences of OmpC, OmpF (Figure 3), and PhoE from E. coli K12 have been determined and found to be 60% homologous (Inokuchi et. al., 1982; Mizuno et. al., 1983). Unlike other membrane proteins, porins have a high percentage of hydrophilic amino acids and lack long stretches of Figure 3. Topology of OmpF, with amino-acid sequence in one- letter code. The view is from outside the 16-stranded antiparallel B-barrel, which is unrolled. The last two 3- stands are repeated on the right-hand side (hatched) to emphasize the barrel. Secondary structural elements are indicated (bold if their side chains are external) by diamonds for barrel B-strands, rectangles for a-helices and circles for turns and loops. The subunit contact regions are shaded. 10 - - - - - - =======- 3 36 3 W I'H‘463 " . r 9 a: 3 oeeelflfilflflfle I oooOGOo 9°09 k 39 ’ . Q9": . . 1‘ °oe I 9 :a‘w'éax‘O- 05% T 1.6"“. \k 1:33;,0 " m *3 ‘zIzre'gwc‘ I .9...” *' g " — °e a" 909° 9am “$80 338 g In ~ 0‘7 Q o "99'. 0 WW » g) -: ME” 9 o ‘9 a 9 " “- C Q 99 ¢ 0 '70 ~ : 0909996 0 99 99.5906 +- co 09990000 (,6 6° 9 00.0 :3 eeooooa 9 '.“‘<’°¢ 9°90 NWOQ ‘99"9 ©9?° 0’9 00 .Ioec’ e°°o°¢§~9e 9 0 9" o 0 Go 6900 o 6 In 2": 19 (.960 696 _, 9909066 99 9.,9 h 9 ' Q I h 69“,, .:. ”fen ,g‘V' 9 0 co ll hydrophobic residues (Tommassen, 1988). Studies using circular dichroism spectroscopy (Markovic-Housley, 1986), infrared spectroscopy (Kleffel et. al., 1985; Nabedryk et. al., 1988), and Raman spectroscopy (Vogel & Jahnig, 1986) have shown that porin‘s secondary structure is high in beta sheet. These observations have lead to two proposals regarding the structure of porin in the membrane: 1) the observed amphipathic character of amino acid sequences 9-10 residues long, which are thought to be transmembrane B- strands suggests the alignment of the hydrophobic sides to face toward the lipid bilayer with the polar side facing the inside of the pore (Tommassen, 1988; Vogel & Jahnig, 1986), and 2) the presence of polar and ionizable residues within the membrane may indicate networks of hydrogen and/or ionic bonds (inside the channel lumen, between subunits, and between headgroups of the membrane lipid and the porin exterior.) (Vogel & Jahnig, 1986; Rosenbusch, 1990). The recent structural analysis of crystallized OmpF to a resolution of 2.4 A has corroborated this model and permitted a more detailed description of porin structure (Figure 4) (Cowan et. al., 1992). Porin is a trimer of ~36 kDa monomers which bind ionically to peptidoglycan as well as interacting strongly with LPS. OmpC, OmpF and PhoE monomers when synthesized simultaneously have been shown to randomly associate into heterotrimers (Gehring & Nikaido, 1989). Each monomer consisting of a 16-stranded antiparallel B-barrel forms a pore or channel (Cowan et. 12 Figure 4. RIBBON diagram of OmpF. Arrows represent B-strands and are labelled 1 to 16 starting from the strand after the first short turn. The short B-strand at the N-terminus continues the C-terminal strand 3-16 and is therefore called 316'. The long loops are denoted L1 to L8, the short turns at the other end T1 to T8. Loop L2 protudes towards the viewer. Loop L3 (black) folds inside the barrel. 13 Figure 4 14 al., 1992). The exterior surface of the trimer complex contains a hydrophobic band of R groups whose upper and lower limits consist of aromatic amino acids, with small size aliphatic residues in between. The Iinterface between monomers is a combination of hydrophobic and hydrophilic interactions. Loops (connecting the B- strands) on the cell exterior, in general, are longer than those facing the periplasm; thus the exterior of the protein has a "rough" surface. On this surface are a number of carboxyl groups which could presumably bind to the anionic LPS via divalent cations. One large loop between B-strand 5 and 6 (L3) folds into the channel lumen approximately midway in the channel interior creating an eyelet and forming a constricted zone. This zone determines the exclusion limit of the pore and is segregated into negatively-charged amino acids on one side within the loop (Asp113 and Glu117) and positively-charged amino acids across from the loop (Arg42, Arg82, Arg132, Lysls). The amino acids in this constriction zone are highly conserved among OmpF, PhoE and OmpC. In terms of interactions between porin and other membrane components, LPS seems to be important in the association of porin into trimers as well as its assembly into the outer membrane (Ried et. al., 1990; Sen & Nikaido, 1991). The isolated trimer is associated with variable amount of LPS as illustrated by the ladder of bands of trimers with different levels of associated LPS seen on a SDS-polyacrylamide gel (Rocque et. al., 1987). One result 15 of the unique structure of porin is its stability to various denaturing conditions such as high temperatures, pH extremes, ionic detergents, and urea (Schindler & Rosenbusch, 1984; Rocque & McGroarty, 1990; Rosenbusch, 1974; Nakae et. al., 1979). Bacterial Porin: In vitro Functional Analyses Various methods have been used to analyze porin function in vitro and in vivo. Two commonly used in vitro techniques are the bilayer lipid membrane (BLM) assay (Schindler & Rosenbusch, 1978; Benz & Bauer, 1988; Benz et. al., 1979) and the liposome swelling assay (LSA) (Nikaido & Rosenberg, 1983). Using the BLM, conductance is measured across a lipid membrane painted across a small hole in a teflon partition separating two chambers. These chambers contain a salt solution with a small amount of porin dissolved in a detergent. Insertion of porin into the membrane results in a measurable increase in membrane conductance. The conductance measured can indicate the insertion and opening of channels as well as their closing. Parameters that can be measured using the BLM assay include the cross-sectional area of the porin channel at its narrowest point, cooperativity of channel opening and closing and ion selectivity. Cross-sectional area is calculated by dividing the specific channel conductance increment, A, by the specific conductance of the bathing solution, a. Channel cooperativity is defined as the 16 opening or closing of the 3 porin monomers as a unit indicated by the conductance levels (trimer channel size). Noncooperativity of channel opening or closing is indicated by channel sizes 1/3 or 2/3 the size of the trimer (Xu et. al., 1986). Effects of various environmental factors that have been examined include pH, voltage, temperature, as well as the presence of LPS. It has been shown that ion flow through the porin channels is similar to movement in free solution (ie. the average k/a is fairly constant over a range of salt concentrations) (Benz & Bauer, 1988; Benz et. al., 1978; Brunen et. al., 1992) although small variations in conductance which have been measured may be due to interaction of carboxyl groups in the pore with cations since amidation of carboxyl groups significantly reduced the variability of k/a (Benz et. al., 1984). Limitations of this method are due to: 1) difficulties in ascertaining which conductance increment represents trimer opening, 2) the fact that the conductance increment is a concerted effect of surface potentials (LPS or phospholipid), charge of porin trimers, and potential difference across the membrane (Brunen et. al., 1992; Jap, 1989), and 3) the fact that the permeability only of ions or charged molecules can be measured and not uncharged solutes. LSA is a technique whereby the channel exclusion limit for uncharged solutes can be determined (Nakae, 1986; Lugtenberg & Alphen, 1983). 'Liposomes are formed in the presence of porin in a solution containing an impermeant 17 solute. The rate of change in the light scattering of such liposomes which swell upon dilution into a solution containing a specific-sized test solute reflects the relative size of the channel. Limitations of this method include the following: 1) the Donnan potential of the intact cell is not reproduced in liposomes and 2) if charged molecules are present, a membrane potential can be created, inducing a complex movement of buffer and other ions (Bellido et. al., 1991). Using BLM analysis, porin has been shown to exist in both an open and a closed channel conformation but the significance of the closed conformation has been questioned. Some investigators have observed an increase in the number of channel closings with increased transmembrane voltage, a phenomenon referred to as voltage gating (Schindler & Rosenbusch, 1978; Schindler & Rosenbusch, 1981; Dargent et. al., 1986; Xu et. al., 1986; Delcour et. al., 1989). Others have been unable to detect this gating phenomenon (Benz, 1985; Benz et. al., 1978; Lakey et. al., 1985). In a recent paper, Lakey and coworkers detected voltage gating using porin isolated and analyzed applying a variety of methods; they suggested that differences in measuring voltage gating may be due to variations in the BLM technique (Lakey & Pattus, 1989). The physiological significance of voltage gating has been questioned because: 1) gating reportedly occurs at voltages (>100 mV) (Schindler & Rosenbusch, 1978; Dargent et. al., 1986; Xu et. al., 1986) beyond that of the 18 outer membrane’s Donnan potential of 30 mv (Stock et. al., 1977) and 2) Sen and coworkers have found that porin channel permeability in intact cells is not affected by high Donnan potentials induced across the outer membrane (Sen et. al., 1988). However acid pH, as is found in the periplasm (Stock et. al., 1977), may reduce the voltage threshold in vivo as it has been found to do in vitro (Xu et. al., 1986). Also a change in electrostatic field strength can be amplified across the narrow porin channels (Itoh & Nishimura, 1986; Jap, 1989) which may allow short term potentials of up to 130 mV to occur at low salt concentrations (Lakey, 1987). Alternatively, Lakey and Pattus (1989) have suggested that _ the removal of the porin-peptidoglycan linkage, occurring as a result of porin purification, may induce voltage dependence and this voltage control may be an expression of an in vivo closing regulated by another mechanism. Others have suggested that voltage gating is a protection to prevent porin, if mistakenly inserted into the cytoplasmic membrane, from collapsing the transmembrane potential (Nikaido, 1992). In addition to affecting gating, pH has been found to influence porin channel size (Benz et. al., 1979; Buehler et. al., 1991) and cooperativity (Xu et. al., 1986). Buehler and coworkers (1991) have observed an increase in channel size at low pH (pH 4.5). On the other hand, Benz and coworkers (1979) detected geereeeeg channel size (the most frequently observed conductance, A, halved) with pH 19 decreases from 9 to 2. Also Schindler and Rosenbusch (1982) observed "very high conductance levels" at pH values above pH 9.5. Non-bacterial ion channels, including the mitochondrial voltage dependent anion channel (VDAC) and .sodium channels, have shown changes in channel function with pH similar to those seen in Benz's studies of bacterial porins. VDAC switches to substates of lower permeability with increasing voltage (Benz, 1990; Mannella, 1990); at acidic pH, the conductance of VDAC has been shown to decrease more sharply (as compared to basic or neutral pH) as membrane potential increases (Ermishkin & Mirzabekov, 1990). With batrachotoxin-modified brain sodium channels, single channel conductance was found to decrease by 50% when the pH was changed from 7.4 to 4.9 (P. Daumas and 0.8. Anderson, Abstr. Annu. Meet. Biophys. Soc. 1991, p. 259). Changes in function, which are pH-dependent, have been observed in other transport systems. For example, the system A transport of alanine in hepatocytes is reported to be inactivated at pH 6.0 and to reach maximum activity at pH 8.0; it has been proposed that the titration of a histidine is involved with this pH-dependent change in activity. (Bertran et. al., 1991). Xu and coworkers detected decreased channel cooperativity at low pH (Xu et. al., 1986). Voltage also appears to control channel cooperativity. PhoE and OmpF show a loss of cooperativity at high potentials as indicated by predominance of monomer and dimer-sized conductances under these conditions 20 (Schindler & Rosenbusch, 1981; Schindler & Rosenbusch, 1978; Xu et. al., 1986). Most investigators have found that bound LPS can modify bacterial porin function (Schindler & Rosenbusch, 1981; Schindler & Rosenbusch, 1978) while some have found that LPS removal has no effect (Benz et. al., 1978; Hancock, 1987). It has been shown that methods used to deplete porin of LPS do not totally remove LPS (Rocque et. al., 1987,) and this could explain why some investigators do not detect an LPS-dependent effect. The anionic MDO has also been found to affect porin function. Declour and coworkers, using patch clamping techniques found that with polarities opposite to the Donnan potential, MDO increased the number of closings of porin from E. coli wild type AW740 (Delcour et. al., 1992). They found that phosphoethanolamine (a substituent of M00) mimicked this effect and hypothesized that the depolarization of cells adapted to low osmolarity, which occurred when cells were suddenly placed in high-salt medium, caused the MOO-dependent closure of porin channels. Similarly polyanions have been found to affect VDAC function; Konig’s polyanion has been found to induce VDAC to close at lower potentials (Mannella, 1992). Colombini and coworkers have isolated a modulator protein from mitochondria which has the same effect as polyanions (Liu & Colombini, 1992). 21 Another factor which has been found to affect function for some E.coli channels is the tension of the membrane lipid. Specifically, Martinac and coworkers have discovered channels in the outer membrane (and in the inner membrane) which can be activated by suction or pressure (Martinac et. al., 1990). These "mechanosensitive" channels were also activated by amphipaths (Markin & Martinac, 1991). This presumably occurred via a mechanism, first described by Sheetz and Singer (1974), whereby the insertion of an amphipath preferentially into one leaflet of an asymmetrical bilayer (either due to the asymmetric distribution of proteins or polar lipids), caused an alteration of the contour of the membrane, ie. a "cupping" effect. Activity of mechanosensitive channels was also increased in the presence of increased osmolarity. These results in combination prompted Martinac and coworkers (1990) to propose the presence of mechanosensitive biosensors in the living cell which monitor membrane expansion during cell growth and volume or osmolarity changes. Bacterial Porin: In vivo Functional Analyaaa Methods used to study porin function in vivo include the measurement of uptake of metabolizable solutes such as antibiotics or maltose. Since 6-lactamases which degrade B- lactams are located primarily in the periplasm, Zimmermann and Rosselet (1977) have developed a method to measure outer membrane permeability by monitoring the decrease of B-lactam 22 (measured as a decrease in absorbance at 260 nm) using intact cells. Assuming that the B-lactams penetrate the outer membrane primarily through porins and that the Km and V for fl-lactamase is the same for cell-bound and mu periplasmic enzyme, they proposed that at a given B-lactam concentration, a steady state is rapidly established such that: C(S. " S.) = S. X Vm/(S. + K...) where C = permeability parameter, SD==external antibiotic concentration, and S===periplasmic antibiotic concentration (Zimmermann & Rosselet, 1977). These assumptions were found to be accurate, at least in experiments with E. coli (Zimmermann & Rosselet, 1977; Sawai et. al., 1977; Nikaido et. al., 1983), allowing the prediction of minimum inhibitory concentrations (MICs) (Livermore & Davy, 1991). Using this method, Nikaido and coworkers (1983) showed that OmpC and OmpF excluded large hydrophobic and negatively- charged solutes. This may help explain how these bacteria protect themselves against hydrophobic and negatively- charged bile salts in their natural environment, the intestinal tract (Nikaido, et. al., 1983). They also found using hydrophilic compounds that OmpC-containing cells had consistantly lower permeability coefficients than OmpF- containing cells, reflecting OmpC's relatively smaller channels. Using whole cells, investigators have observed that acidic pH reduces the effectiveness of certain hydrophilic 23 antibiotics as measured by the MIC, bacteriocidal and post- antibiotic effects (suppression of bacterial growth which persists after limited exposure of organism to antimicrobial agent) (Gudmundsson et. al., 1991; Laub et. al., 1989; Sabath et. al., 1968). It has been suggested that this was due not to the altered ionization state of the antibiotic but to a change in "receptor" (ie. porin) affinity on the bacterial cell (Sabath et. al., 1968). Bacterial Porin: Structure-Function studies A variety of approaches have been used to define regions of porin structure that are vital for function. Chemical modification (Tokunaga et. al., 1981; Benz et. al., 1984; Hancock et. al., 1986; Schindler & Rosenbusch, 1982), mutation (Misra & Benson, 1988a; Misra & Benson, 1988b), and hybrid porin gene construction (Tommassen et. al., 1985; Benz et. al., 1989; van der Ley et. al., 1987; Nogami et. al., 1985) experiments have been utilized to determine which amino acids are important in defining channel function. Chemical modification of amino and carboxyl groups in porin has been found to slow the diffusion of negatively and positively charged molecules respectively; neither of these modifications has been shown to affect diffusion of uncharged solutes (Tokunaga, 1981). Chemical modification has also been used to analyze exposure of amino acids and to probe conformational changes (Schindler & Rosenbusch, 1984; Schindler & Rosenbusch, 1982). Chemical modification of 24 lysine with eosin isothiocyanate was shown to increase with increasing pH (Schindler 8 Rosenbusch, 1984); these workers suggested that the change in modification with pH resulted from a pH-dependent conformational change in porin rather 'than solely from increased protonation of amino groups at basic pH. Benson and coworkers have isolated strains of E. coli producing mutant OmpC and OmpF porins which allow the cells to grow on maltodextrins in the absence of the maltodextrin- specific porin LamB (Dex+ phenotype) (Misra 8 Benson, 1988a; Misra 8 Benson, 1988b). These porins apparently have enlarged channels and the change in channel structure in these Dex+ strains caused an increased sensitivity to hydrophilic antibiotics as well as an increased rate of 1‘C- maltose uptake in the intact cells. The mutant porins contained amino acid deletions, insertions and single residue substitutions. Benson observed that these mutations all occurred in the first third of the protein and involved a small number of charged amino acids (R42, R82, R132, 0113 for OmpF) (Misra 8 Benson, 1988a, Misra 8 Benson, 1988b). According to the most recent structural analysis of OmpF, these amino acids contribute to the eyelet of the channel; changes in the length or charge of these amino acids can change the exclusion limit of the porin channel (Cowan et. al., 1992). 25 Bacterial Porin: significance of structure-Function studies There are three major reasons for looking at structure-function relationships in bacterial porin. First, such studies may help discover ways to increase the antibiotic suseptibility of pathogenic bacteria. For example, Sabath and coworkers indicated that alkalinization of the media had an enhancing effect on the action of erythromycin against gram-negative bacteria (Sabath et. al., 1985). The second reason for studying porin structure- function relationships is as a model for other channel- forming proteins. As described previously, mitochondrial porins are similar to bacterial porins in function as well as structure (Ermishkin 8 Mirzabekov, 1990; Blachly-Dyson et. al., 1990). However bacterial porins are more amenable to study due to the ease of genetic manipulation as well as to the level of protein that can be isolated. Also the primary sequence and three dimensional structure of bacterial porin is more defined. Finally, there has recently been a study showing that porins play a role as pathogenicity determinants in gram-negative bacterial infections (Galdiero et. al., 1990; Li et. al., 1991). Thus elucidation of porin structure-function relationships could be a step toward altering pathogenicity. REFERENCES Bellido, P., Pechere, H. J., 8 Hancock, R. E. W. (1991) Antimicrob. Ag. Chemother. 35, 68-72. Benz, R. (1990) Experientia 46, 131-136. Benz, R. (1985) CRC Rev. Biochem. 19, 145-190. Benz, R., 8 Bauer, K. (1988) Eur. J. Biochem. 176, 1-19. Benz, R., Janko, K., Boos, w., 8 Lauger, P. (1978) Biochim. Biophys. Acta 511, 305-319. Benz, R., Janko, R., 8 Lauger, P. (1979) Biochim. Biophys. Acta 551, 238-247. Benz, R., Schmidt, A., van der Ley, P., 8 Tommassen, J. (1989) Biochim. Biophys. Acta 981, 8-14. Benz, R., Tokunaga, H., 8 Nakae, T. (1984) Biochim. Biophys. Acta 769, 348-356. Blachly-Dyson, B., Peng, S., Colombini, M., 8 Forte, M. (1990) Science 247, 1233-1236. Buehler, L. K., Kusumoto, S., Zhang, H., 8 Rosenbusch, J. P. (1991) J. Biol. Chem. 266, 24460-24450. Brunen, M., Engelhardt, H., Schmidt, A., 8 Benz, R. (1992) J. Bacteriol. 173, 4182-4187. Chang, C. P., Mizushima, S., 8 Glaeser, R. M. (1985) Biophys. J. 47, 629-639. Cowan, S. W., Schirmer, T., Rummel, G., Steiert, M., Ghosh, R., Pauptit, R._A.; Jansonius, J. N., 8 Rosenbusch, J. P. (1992) Nature 358, 727-733. Dargent, B., Hofmann, W., Pattus, F., 8 Rosenbusch, J. P. (1986) EMBO J. 5, 773-778. Dorset, D. L., Engel, A., Massalski, A., 8 Rosenbusch, J. P. (1984) Biophys. J. 45, 128-129. Ermishkin, L. N., 8 Mirzabekov, R. A. (1990) Biochim. Biophys. Acta 1021, 161-168. Fourel, D., Mizushima, S., 8 Pages, J. (1991) Eur. J. Biochem. 206, 109-114. 26 27 Galdiero, F., Tufano, M. A., Galdiero, M., Masiello, S., 8 DiRosa, M. (1990) Infection and Immunity 58, 3183-3186. Gehring, K. B., 8 Nikaido, H. (1989) J. Biol. Chem. 264, 2810-2815. Gudmundsson, A., Erlendsdottir, H., Gottfredsson, M., 8 Gudmundsson, S. (1991) Antimicrob. Ag. Chemother. 35, 2617- 2624. Hancock, R. E. W. (1987) J. Bacteriol. 169, 929-933. Hancock, R. E. W., Schmidt, A., Bauer, K., 8 Benz, R. (1986) Biochim. Biophys. Acta 860, 263-267. Heyde, M., and Portalier, R. (1987) Mol. Gen. Genet. 208, 511-517. Inokuchi, K., Mutoh, N., Matsuyama, S., 8 Mizushima, S. (1982) Nucleic Acids Research 10, 6957-6968. Itoh, S., 8 Nishimura, M. (1986) Math. Enz. 125, 58-86. Jap, B. K. (1989) J. M01. Biol. 205, 407-419. Jap, B. K., Downing, K. H., 8 Walian, P. J. (1990) Structural Biology 103, 57-63. Jap, B. K., Walian, P. J., 8 Gehring, K. (1991) Nature 350, 167-170. Jeanteur, D., Lakey, J. H., 8 Pattus, F. (1991) Mol. Microbiol. 5, 2153-2164. Kennedy, E. P. (1982) Proc. Natl. Acad. Sci. USA 79, 1092- 1095. Kleffel, B., Garavito, R. M., Baumeister, W., 8 Rosenbusch, J. p. (1985) 4, 1589-1592. Lakey, J. H., 8 Pattus, F. (1989) Eur. J. Biochem. 186, 303- 308. Lakey, J. H., Watts, J. P., 8 Lea, E. J. A. (1985) Biochim. Biophys. Acta 817, 208-216. Laub, R., Schneider, Y., 8 Trouet, A. (1989) J. Gen. Microb. 135, 1407-1416. Id, 2. M., Hannah, J. H., Stibitz, S., Nguyen, N. Y., Manclark, C. R., 8 Brennan, M. J. (1991) Mol. Microbiol. 5, 1649-1656. 28 Liu, M. Y., 8 Colombini, M. A. (1992) J. Bioenerg. Biomembr. 24, 41-46. Livermore, D. M., 8 Davy, K. W. M. (1991) Antimicrobiol. Ag. Chemother. 35, 916-921. Lugtenberg, B., 8 Alphen, L. V. (1983) Biochim. Biophys. Acta 737, 51-115. Mannella, C. A. (1990) Experientia 46, 137-145. Mannella, C. A., Forte, M., 8 Colombini, M. (1992) J. Bioenerg. Biomembr. 24, 7-19. Markin, V. S., 8 Martinac, B. (1991) Biophys. J. 60, 1-8. Markovic-Housley, 2., 8 Garavito, M. (1986) Biochim. Biophys. Acta 869, 158-170. Martinac, B., Delcour, A. H., Buechner, M., Adler, J., 8 Kung, C. (1990) in Comparative Aspects of Mechanoreceptor Systems (Fumio Ito, ed.) Springer-Verlag, New York. Meyer, 8. B., Granett, S., Jung, J. U., 8 Villarejo, M. R. (1990) J. Bacteriol. 172, 5501-5502. Misra, R., 8 Benson, S. A. (1988a) J. Bacteriol. 170, 3611- 3617. Misra, R., 8 Benson, S. A. (1988b) J. Bacteriol. 170, 528- 533. Mizuno, R., Chou, M., 8 Inouye, M. A. (1983) J. Biol. Chem. 258, 6932-6940. Nabedryk, E., Garavito, R. M., 8 Breton, J. (1988) Biophys. J. 53, 671-676. Nakae, T. (1986) CRC Crit. Rev. Microbiol. 13, 1-62. Nakae, T., Ishii, J., 8_Tokunaga, M. (1979) J. Biol. Chem. 254, 1457-1461. Nikaido, H. (1992) Mol. Microbiol. 6, 435-442. Nikaido, H. 8 Rosenberg, E. Y. (1983) J. Bacteriol. 153, 241-252. Nikaido, H., Rosenberg, E. Y., 8 Foulds, J. (1983) J. Bacteriol. 153, 232-240. Nikaido, H., 8 Vaara, M. (1985) Microbiol. Rev. 49, 1-32. 29 Nogami, T., Mizuno, T., 8 Mizushima, S. (1985) J. Bacteriol. 164, 797-801. Paul, C., 8 Rosenbusch, J. P. (1985) EMBO J. 4, 1593-1597. Reid, J., Fung, H., Gehring, K., Klebba, P. E., 8 Nikaido, H., (1988) J. Biol. Chem. 263, 7753-7759. Ried, G., Hindennach, I., 8 Henning, U. (1990) J. Bacteriol. 172, 6048-6053. Rocque, W. J., Coughlin, R. T., 8 McGroarty, E. J. (1987) J. Bacteriol. 169, 4003-4010. Rocque, W. J., 8 McGroarty, E. J. (1990) Biochemistry 29, 5344-5351. Rosenbusch, J. P. (1974) J. Biol. Chem. 249, 8019-8029. Rosenbusch, J. P. (1990) Experientia 46, 167-173. Sabath, L. D., Lorian, V., Gerstein, D., Loder, P. B., 8 Finland, M. (1968) Appl. Microbiol. 16, 1288-1292. Sawai, T., Matsuba, K., 8 Yamagishi, S. (1977) J. Antibiotics 30, 1134-1136. Schindler, H., 8 Rosenbusch, J. P. (1978) Proc. Natl. Acad. Sci. USA 75, 3751-3755. Schindler, H., 8 Rosenbusch, J. P. (1981) Proc. Natl. Acad. Sci. USA 78, 2302-2306. Schindler, M., 8 Rosenbusch, J. P. (1982) J. Cell. Biol. 92, 742-746. Schindler, M., 8 Rosenbusch, J. P. (1984) FEBS Lett. 173, 85-89. Schulein, K., 8 Benz, R. (1990) Mol. Microbiol. 4, 625-632. Sen, K., Hellman, J., 8 Nikaido, H. (1988) J. Biol. Chem. 263, 1182-1187. Sen, K., 8 Nikaido, H. (1991) J. Bacteriol. 173, 926-928. Sheetz, M. P., 8 Singer, S. J. (1974) Proc. Natl. Acad. Sci USA 71, 4457-4461. Stock, J. B., Rauch, B., 8 Roseman, S. (1977) J. Biol. Chem. 252, 7850-7861. 30 Tokunaga, H., Tokunaga, M., 8 Nakae, T. (1981) J. Biol. Chem. 256, 8024-8029. Tommassen, J. (1988) in Membrane Biogenesis (J. A. F. 0p den Kamp, ed.) pp. 353-373 Springer-Verlag, New York. Tommassen, J., van der Ley, P., van Zeijl, M., 8 Agterberg, M. (1985) EMBO J. 4, 1583-1587. van der Ley, P., Burm, P., Agterberg, M., van Meersbergen, J., 8 Tommassen, J. (1987) Mol. Gen. Genet. 209, 585-591. Van Golde, L. M. G., Schulman, H., 8 Kennedy, E. P. (1973) Proc. Natl. Sci. USA 70, 1368-1372. Vogel, H., 8 Jahnig, F. (1986) J. Mol. Biol. 190, 191-199. Weiss, M. S., Abele, U., Weckesser, J., Welte, W., Schiltz, B., Schulz, G. E. (1991) Science 254, 1627-1630. Xu, 6., Shi, B., McGroarty, E. J., Tien, H. T. (1986) Biochim. Biophys. Acta 862, 57-64. Zimmermann, W., 8 Rosselet, A. (1977) Antimicrob. Ag. Chemother. 12, 368-372. CHAPTER 2 Effects of pH on Bacterial Porin Function 31 ABSTRACT Porin is a trimeric channel-forming protein in the outer membrane of gram-negative bacteria. Function of the porins OmpF, OmpC and PhoE from Escherichia coli K12 were analyzed at various pHs. Preliminary results from bilayer lipid membrane and liposome swelling assays indicated that in vitro, porin has at least two open channel configurations with a small and large size. The small channels were stabilized at low pH while the larger channels were detected under basic conditions. The size-switch occurred over a very narrow range near neutral pH, and the two major open channel configurations responded differently to variations in voltage. The presence of two or more pH-dependent substates of porin could explain the variability in pore diameter measured by others and suggests a more dynamic role for porin in the cell. 32 INTRODUCTION Diffusion of small hydrophilic molecules (5 600 Daltons) through the outer membrane of gram-negative “bacteria occurs via pore-forming proteins called porins. The two main non-specific porins of Escherichia coli are OmpC and OmpF. OmpF has a larger channel than OmpC and is induced under conditions of low osmolarity. PhoE is an anion-selective porin whose production is induced by phosphate limitation. Recent studies have indicated that PhoE synthesis is also inhibited by high osmolarity (Meyer et. al., 1990). Because porins play such an important role in solute uptake, their structures have been examined extensively. The primary sequences of OmpC, OmpF, and PhoE from E. coli K12 have been determined and have been found to be 60% homologous (Inokuchi et. al., 1986; Mizuno et. al., 1983). Unlike other membrane proteins, porins have a high percentage of hydrophilic amino acids and lack long stretches of hydrophobic residues (Tommassen, 1988). Studies using various spectroscopic techniques have shown that porin's secondary structure is high in B-sheet (Kleffel et. al., 1985; Markovic-Housley 8 Garavito, 1986; Nabedryk et. al., 1988; Vogel 8 Jahnig, 1986). Such studies have led to two proposals regarding the tertiary structure of porin in the membrane: (1) The amphipathic character of amino acid sequences 9-10 residues long, thought to be 33 34 transmembrane B-strands, suggests an alignment of the hydrophobic sides toward the lipid bilayer with the polar sides facing the inside of the pore (Tommassen, 1988; Vogel 8 Jahnig, 1986). (2) The presence of polar and ionizable residues within the membrane may indicate networks of hydrogen and/or ionic bonds (Paul 8 Rosenbusch, 1985; Rosenbusch, 1990). Ionic bond formation could allow for pH- dependent structural changes. In terms of quaternary structure, porin is a trimer of ~36 kDa monomers. The three-dimensional structure of PhoE, OmpC, and OmpF has been examined by electron microscopic image reconstruction (Chang et. al., 1985; Dorset et. al., 1984; Jap et. al., 1990; Jap et. al., 1991). Using such a technique, Jap and coworkers analyzed PhoE and found that the channels in the trimer complex converge as they traverse the membrane but do not merge (Jap et. al., 1990). Cowan and coworkers' (1992) recent x-ray diffraction analysis of crystallized OmpF at a resolution of 2.4 A has allowed more detailed analysis of porin. Each monomer consists of a 16-stranded antiparallel B-barrel containing a pore (Cowan et. al., 1992). The exterior surface contains a hydrophobic band whose upper and lower limits consist of aromatic amino acids, with small size aliphatic residues in between these limits. The interface between monomers contains a combination of hydrophobic and hydrophilic interactions. The loops (connecting the B-strands) facing the cell exterior in general are longer than those facing the periplasm and thus 35 create a "rough" outer surface. On this surface are carboxyl groups which could presumably bind to LPS via divalent cations. One large loop between B-strand 5 and 6 folds into the channel luman about half-way down creating an eyelet or a constricted zone. This zone determines the exclusion limit of the pore and is segregated into negatively-charged amino acids on one side (Asp113 and Glu117) and positively-charged amino acids on the other (Arg42, Arg82, Argl32, Lys16). The amino acids in this constriction site are conserved among OmpF, PhoE, and OmpC. Previous studies have suggested that LPS plays a vital role in the structural integrity of porin (Hoenger et. al., 1990; Rocque et. al., 1987; Xu et. al., 1986) and is critical for trimerization (Sen 8 Nikaido, 1991), insertion (Ried et. al., 1990) and functioning (Schindler 8 Rosenbusch, 1981) of porin in the outer membrane. As a result of porin’s structure as well as its strong interactions with LPS, the trimeric complex is stable to various denaturing conditions such as high temperatures, pH extremes, ionic detergents, and urea (Nakae et. al., 1979; Rocque 8 McGroarty, 1990; Rosenbusch, 1974; Schindler 8 Rosenbusch, 1984). Structural studies have been complemented by functional analysis of porin. Using the bilayer lipid membrane (BLM) assay, porin has been shown to exist in both an open and a closed channel conformation but the significance of the closed conformation has been questioned (Benz, 1985; Benz 36 et. al., 1982; Hancock, 1987; Lakey et. al., 1985, Sen et. al., 1988). In a recent study, Lakey 8 Pattus (1989) detected voltage-induced channel closing (ie., gating) using porin isolated by various techniques and modifying the BLM protocol; the results suggested that discrepancies among various investigators concerning voltage gating may be due to differences in BLM analysis techniques. The physiological significance of voltage gating has been questioned because: 1) gating seems to occur at voltages (>100 mV) beyond that of the outer membrane's Donnan potential of 30 mV (Benz et. al., 1982; Dargent et. al., 1986; Schindler 8 Rosenbusch, 1978; Xu et. al., 1986) and 2) porin channel permeability in intact cells is not affected by high Donnan potentials induced across the outer membrane (Sen et. al., 1988). However, Stock et. a1. (1977) have suggested that the pH in the periplasm may be significantly below that in the media; such a lowered pH in vivo may decrease the threshold for voltage gating similar to the decrease detected in vitro at acidic pH (Xu et. al., 1986). Also the strength of the electric field in a narrow channel can be greater than that across the membrane (Itoh 8 Nishimura, 1986; Jap, 1989) and may allow for short term potentials of up to 130 mv to occur at low salt concentration (Lakey, 1987). In addition to affecting gating, pH has been found to influence the cooperativity (defined as the 3 monomers opening simultaneously in the trimer unit) of E. coli porin 37 channels (Xu et. al., 1986) and to affect channel size (Benz et. al., 1979; Schindler 8 Rosenbusch, 1978). Xu and coworkers (1986) observed a decrease in E. coli porin channel cooperativity at low pH. Benz and coworkers (1979) found an increase in channel size with pH. Similarly, Schindler and Rosenbusch (1978) observed "very high conductance levels" at pH values above pH 9.5. Also, Schindler and Rosenbusch (1982) observed increased modification of lysine with eosin isothiocyanate when the derivitization of porin was carried out at higher pH. Based on their experiments, they suggested that this increased modification resulted from a pH-dependent conformational change in porin rather than solely from decreased protonation of amino groups. Non-bacterial ion channels, including the mitochondrial voltage dependent anion channel (VDAC) and sodium channels, have shown similar changes in channel function with pH. For example, at acidic pH, the conductance of VDAC has been shown to decrease sharply as membrane potential increases; while the decrease in conductance with voltage is not as great at basic pH (Ermishkin 8 Mirzabekov,_1990; K. W. Kinally, Abstr. Annu. Meet. Biophys. Soc. 1991, p. 177). With batrachotoxin- modified brain sodium channels, single channel conductance was found to decrease by 50% when the pH was changed from 7.4 to 4.9 (P. Daumas and 0. S. Anderson, Abstr. Annu. Meet. Biophys. Soc. 1991, p. 259). The functional properties of other transport systems have also been shown to be pH- 38 dependent. The system A transport of alanine in hepatocytes is reported to be inactivated at pH 6.0 and to reach maximum activity at pH 8.0; it has been proposed that the titration of a histidine is involved with this pH-dependent change in function (Bertran et. al., 1991). In this study, we examined the effects of pH on E. coli K12 porin function. We have observed at least two main conformations of the open channel, a small-size channel, stable at acidic pH, and larger-sized channels, stable at basic pH. We found that the pH-induced switch in channel size occurred in the neutral pH range and was affected by high voltage and, in some cases, by LPS depletion. The mechanism of the pH-induced changes and the possible physiological role of pH-induced regulation of porin channel activity in the cell are discussed. MATERIALS AND METHODS Cell Growth OmpC was isolated from E. coli strain ECB 621 which lacks OmpF and LamB (gift of S. Benson). OmpF and PhoE were isolated from E. coli K12 strains PLB 3261 (ompC‘, lamB’, Benson 8 Decloux, 1985) and JF 694 (OmprOmpCV gift of S. Benson) respectively. The strains were grown in 1% tryptone, 0.5% yeast extract, and 0.4% NaCl, pH 7.5 as described previously (Rocque et. al., 1987; Rocque 8 McGroarty, 1989; Rocque 8 McGroarty, 1990). Cells producing OmpC and PhoE were grown at 37°C, and cells synthesizing OmpF were grown at BUTn Cells were harvested late in logarithmic growth phase. Porin Isolation Porins were isolated by the method of Lakey et. al. (1985) with some modifications (Rocque et. al., 1987; Rocque 8 McGroarty, 1989, Rocque 8 McGroarty, 1990). Cells were broken using a French press and treated with RNAse and DNAse. After pelleting the membranes at 100,000xg, the inner membrane and some outer membrane proteins were dissolved with sodium dodecyl sulfate (SDS) and Tris HCl. After centrifugation, the pellet contained outer membrane proteins, including most of the porin, bound to the peptidoglycan. The porins were solubilized with high NaCl in the presence of mercaptoethanol, Tris-HCl, and SDS as 39 40 described previously. The solubilized porin was dialyzed and precipitated with 90% acetone. Homogeneity of porin was assessed by SDS-polyacrylamide gel electrophoresis (SDS- PAGE) using the method of Laemmli (1970). The final preparations were suspended in 1% SDS, 10 mM Tris-HCl pH 6.8, and 0.02% sodium azide (standard buffer). These "LPS- enriched" porin samples contained 6-9 molecules of LPS/trimer as determined by the thiobarbituric acid assay (Droge et. al., 1970). To deplete porins of LPS, the samples were suspended in 30% SDS, applied to a Sephadex G- 200 gel filtration column (2.5 x 100 cm) and eluted with 1% SDS, 10 mM Tris, 0.2 M NaCl, 1 mM EDTA, and 0.02% sodium azide, pH 9.0. This procedure removed all detectable LPS (by the thiobarbituric acid assay), but there probably was a small amount of "unremovable" LPS attached, as shown by Rocque et. a1. (1987). After LPS depletion, the porin remained in a trimeric configuration (Rocque et. al., 1987). Porin was quantitated using the bicinchoninic acid protein assay (Pierce Chemical Co.), and porin trimeric structure and LPS association was monitored by SDS-PAGE (Rocque et. al., 1987) and by the thiobarbituric acid assay. The aggregation state of porin was monitored by measuring light scattering (at 320 nm for concentrations from 0.38 to 3.1 mg/ml) of porin solutions at pH 5.4 and 9.4. 41 Bilayer Lipid Membrane Assay (BLM) Electrical conductance across a bilayer lipid membrane was used to measure porin channel size, cooperativity, and voltage gating under various conditions. The electrical conductance was measured across a lipid bilayer comprised of 1% diphytanoyl phosphatidylcholine (in n-decane). As described previously (Rocque 8 McGroarty, 1989; Rocque 8 McGroarty, 1990; Xu et. al., 1986), a small volume of porins suspended in standard buffer was added to the salt solution bathing the lipid membrane. This bathing solution contained 0.5 M NaC1 and 0.5 mM of an appropriate buffer (2-(N- cyclohexylamino)ethanesulfonic acid, CHES, for pH 8.1-9.4; sodium phosphate for pH 6.25-7.9; or sodium succinate for pH 5.4-5.8). It was difficult to stabilize the pH during the time period of each BLM study but the pH was maintained to within 0.1 pH units of the original pH. Silver-silver chloride electrodes were placed on either side of the membrane and a constant voltage was applied using a 1.5V battery. Changes in current were amplified using a Keithley Model 614 electrometer and recorded. The changes in current were reported as a size parameter A/a (channel conductance increment/bathing solution's specific conductance) versus the probability of the occurrence of an event with a particular size. A statistically significant number of channels (2200) were analyzed for each experimental condition. 42 Liposome Swelling Assay (LSA) LSA was performed exactly as described previously (Rocque 8 McGroarty, 1989; Rocque 8 McGroarty, 1990) using the procedures of Nikaido 8 Rosenberg (1983). Solutions of 17% dextran, and 120 mM dextrose, maltose, and maltotriose were prepared at pH 5.4 using 50 mM succinate and at pH 9.4 using 50 mM CHES. Liposomes were formed using ~6.2 umol phosphatidylcholine in the presence of 17% dextran and “20 ug of LPS-enriched porin, either OmpC or OmpF. Control liposomes lacking porin were formed adding only LPS in an amount equivalent to that present in the porin being tested. Liposomes were added to solutions of the various test sugars at pH 5.4 or 9.4, and sugar influx was measured by the decrease in light scattering at 400 nm using a Gilford Response II spectrophotometer. The rate of swelling [d(0D)/dt] was monitored for the first 30 seconds following dilution and reported as percent decrease in light scattering. The maximum measured d(0D)/dt was —0.146 OD,00 units/min. Nikaido and Rosenberg (1983) reported this method to be accurate at swelling rates (measured as d(0D)/dt) of between -0.,005 and —0.200 0D,00 units/min. RESULTS pH Effects on Channel size In this study, we have identified pH-induced changes in Channel size of porins from E. coli K-12. Three porins of E. coli K-12, OmpF, OmpC, and PhoE, were analyzed at different pHs using the BLM system. The resulting plots of stepwise current changes versus time (Figure 1) were used to calculate the ratio, k/a (channel conductance increment/specific conductance of the bulk aqueous phase) for the porins under various conditions (Figure 2). This ratio was used as a size parameter since: 1) the ratio is proportional to the cross-sectional area of the pore at its narrowest point (assuming the pore is a cylinder filled with a solution of the same conductivity as the external solution; Benz 8 Bauer, 1988), 2) this ratio corrects for variations in the conductivity of the bulk aqueous phase (Benz 8 Bauer, 1988) and 3) no assumptions about the thickness of the membrane need to be made. Since the assumptions made in (1) may not be accurate and since the geometry of the channel is not well defined (Nikaido, 1992), this ratio was not used to calculate channel diameters. ngpgriggng of channel size using this parameter are valid since current increments reflect channel cross-sectional areas. The results indicated that OmpF at pH 5.4 has a predominance of small channels (size parameter of ~1.6 A) and very few large channels (size parameter of 43 44 Figure 1. Stepwise current changes across a membrane comprised of phosphatidylcholine in the presence of LPS- depleted OmpF in a bathing solution of 0.5 M NaCl and 0.5 mM succinate (pH 5.4). The current decrements shown by arrows presumably represent channel closing events. The voltage across the membrane was 75 mV. The tracing goes from right to left. 45 5 min 1267 ns Figure 1 46 Figure 2. Probability distribution histograms of the size parameter A/o (A), for LPS-enriched wild type OmpF (A), 0mm: (B), and PhoE (C) as measured in bilayer lipid membranes. The porins, solubilized in 1% SDS, 10 mM Tris, 0.02% sodimm azide, pH 6.8, were added to bathing solutions of 0.5 M NaCl containing 0.5 mM of either sodium succinate (pH 5.4 or 5.8), sodium phosphate (pH 6.25-7.9), or sodium 2-(N- cyclohexylamino)ethanesulfonic acid (CHES, pH 8.1-9.4). Electrical conductance was measured using a transmembrane potential of 25 mV. 1 is the channel conductance increment and a is the specific conductance of the bathing solution. P, in arbitrary units is the relative number of events with a given size parameter range. Both opening and closing events are included in the histogram of 2200 events. 47 OmpF Figure 2 48 ~3.2-3.5 A, see Figure 3). When measured at pH 6.55, this porin shows fewer small channels and an increase in the number of large channels. This trend continued as the pH was increased to 7.5 and 8.15, until at pH 9.4, most 'channels were of the large-size type. This pattern is referred to as a pH-induced switch in channel size (or more correctly cross-sectional area, but "size" will be used for the purpose of discussion). Histograms of OmpC channel sizes at different pH’s showed a similar pattern, with a preponderance of small channels (size parameter of ~0.9 A) occurring at pH 5.4 and a higher proportion of larger channels (size parameter of ~1.9-2.4 A , see Figure 3) being detected as the pH was increased from 6.25 to 6.5, 8.1, and 9.4. This pattern was also repeated with PhoE; at pH 7.9 and 9.2, a high proportion of large channels (size parameter of ~2.4 A) was measured while the level of small channels (size parameter of ~1.3 A) was increased at the lower pH's, e.g. pH 5.8. In all cases, the size of the large channel was approximately twice that of the small channel (for pH’s < 7.5; see below). No differences in light scattering between porin solutions at pH 5.4 and 9.4 were observed (for concentrations up to 13.1 mg/ml) indicating that the pH-induced switch in channel size, measured using BLM, was not a result of alteration in 49 Figure 3. Average size (A/a) of the large channel at pH‘s above 6.5 for LPS-enriched OmpF (A) and OmpC (B). Each point is based on analysis of at least 200 channels at each pH. 50 A .. o 1 1... 9 .. 8 w. I. 7 .i 6 1 1. 1 5. 4. s 3 3 3 as ._mzz<:o . momfi mo mum mo< 16 _ _ _ r 1.. 8. ...... m 2 2 2 z a: ._mzz<:o m0m<.._ ".0 mN_m m0< 51 aggregation state. Also, we are assuming that the current increments of the small and large size channel are the result of conductance across the three channels in the trimer complex as it inserts and opens cooperatively in the BLM (see Discussion). Detailed analysis of the size of the large channel population for both OmpF and OmpC (Figure 3) measured at pH's above 6.5 indicated an increase in size beginning at pH 7.5 and continuing up to pH 9.4. This increase in the size of the large channel with increasing pH was greater for OmpC than for OmpF and suggests either multiple "large channel" substates or a gradual widening the large channel as the pH was increased from 7.5 to 9.4. To determine which amino acid(s) may be involved in the pH-induced switch in channel size, the pK. of the switch was measured using BLM. Analysis of the change in proportion of large channels with pH for OmpC and OmpF (Figure 4) indicated that the switch from small to large channel size occurred over a fairly narrow pH range. This suggests that the change in channel size may be induced by the titration of a single group with an apparent pK. of 7.2 for OmpF and 6.5 for OmpC. Also, the size-switch was found to be reversible when the pH was adjusted between values above and below the pH, of the size-switch (pH 5.6 and 7.4--data not shown). In addition, porins treated at pH values as high as 8.5 were shown to completely switch to the smaller channel size when the pH was lowered (data not shown). We presume 52 Figure 4. Titration curve for the pH-induced switch in channel size for OmpC (A), and OmpF (B), based on bilayer lipid membrane analysis performed as described in Figure 2. The porins, solubilized in 1% SDS, 10 mM Tris, 0.02% sodium azide, pH 6.8, were added to bathing solutions of 0.5 M NaCl containing an appropriate buffer. Electrical conductance was measured using a transmembrane potential of 25 mV. 53 OmpC i _ l _ . _ _ _ _ 6 4 2 O 8 6 4 2 9 8 7. 6 4..o 2 l 52.98.; 3.12310 momfi .\. IO OmpF L . . . . _ . O O 0 O 0 0 7 m 5 4. 3 2 .l AoS) were. isolated and their functions in vitro compared to wild type OmpF's using the BLM assay. We did not find any evidence for enlarged channel size with the mutant porins possibly due to differences in bound LPS which is know to modulate the pH-induced switch in channel size. Our experiments did seem to implicate 0113 in voltage gating since the porin with a mutation in this amino acid showed very little gating (compared to wild type OmpF at 75 mV). In combination, our_structural studies seemed to complement our experiments examining porin function by showing a small change in porin tertiary structure with either an alteration in pH or modification of Hile. With the availabiity of detailed structural analysis of OmpF, changes in structure under various environmental conditions can be examined further using molecular modeling techniques. 180 Further experiments suggested by these functional and structural studies of porin are 1) examining the effect of pH on porin heterotrimers since they are known to exist in vivo, 2) studying the effect of mutation of Hile on porin function and structure, and 3) continuing studies on the effect of such periplasm components as MDO and peptidoglycan on porin function and structure. These studies along with our current data expand our knowledge of porin function and structure allowing further refinement in such fields as antibiotic design, X-ray crystallographyic analysis of porin, and antimicrobial therapy. REFERENCES Benson, 8. A., 8 Decloux, A. (1985) J. Bacteriol. 161, 361- 367. Cowan, S. W., Schirmer, T., Rummel, G., Steiert, M., Ghosh, R., Pauptit, R. A., Jansonius, J. N., Rosenbusch, J. P. (1992) Nature 359, 727-733. Delcour, A. H., Adler, J., Kung, C., 8 Martinac, B. (1992) FEBS Lett 304, 216-220. Gudmundsson, A., Erlendsdottir, H., Gottfredsson, M., 8 Gudmundsson, S. (1991) Antimicrobiol. Ag. Chemother. 35, 2617-2624. Heyde, M., 8 Portalier, R. (1987) Mol. Gen. Genet. 208, 511- 517. Lakey, J. H., 8 Pattus, F. (1989) Eur. J. Biochem. 186, 303- 308. Nikaido, H. (1991) Mol. Microbiol. 6, 435-442. Sheetz, M. P., 8 Singer, 8. J. (1974) Proc. Natl. Acad. Sci. USA 71, 4456-4461. ’ 181 IV. L I! llllll 163 53 IBRQRIES 13E3€5 nICI-IIGnN STATE UN illll“WlllllllllllllmlillWI 31293008